With respect to voice recognition functions installed in car-navigation systems or the like, a passenger (user) generally specifies (or orders) a start of speech for the system. For that purpose, a button for ordering a start of voice recognition (hereinafter, referred to as a voice-recognition-start ordering unit) is displayed on a touch panel or mounted on a steering wheel. Then, the voice is recognized that is spoken after the voice-recognition-start ordering unit is pushed down by the passenger (user). That is, the voice-recognition-start ordering unit outputs a voice-recognition start signal, and a voice recognition unit, upon receiving that signal, detects from voice data acquired by a voice acquisition unit after reception of that signal, a voice period corresponding to the content spoken by the passenger (user) to thereby perform processing of voice recognition.
However, there is also a voice recognition device that always recognizes the content spoken by the passenger (user) even without a voice-recognition start order by the passenger (user). That is, even though not receiving a voice-recognition start signal, the voice recognition unit detects a voice period corresponding to each content spoken by the passenger (user) from voice data acquired by the voice acquisition unit and extracts feature values of the voice data in that voice period, to thereby repeatedly perform: processing for recognition on the basis of the feature values using a recognition dictionary; and processing for outputting a character string resulting from voice recognition or for searching a database on the basis of the character string to display a search result.
In Patent Document 1, for example, there is disclosed a voice recognition device in which voice recognition is performed by constantly inputting a voice spoken by a user, its recognition result is displayed, and thereafter, processing based on the recognition result is executed through a determination operation made by the user using an operation button.